Bibliografische Information der Deutschen Bibliothek Die Deutsche Bibliothek verzeichnet diese Publikation in der deutschen

Nationalbiografie; detaillierte bibliografische Daten sind im Internet über http://dnb.ddb.de abrufbar.

ISBN 978-3-939473-17-6 Impressum Herausgeber: Der Rektor der Technischen Universität Ilmenau Univ.-Prof. Dr. rer. nat. habil. Peter Scharff Redaktion: Referat Marketing und Studentische Angelegenheiten Kongressorganisation

Andrea Schneider Tel.: +49 3677 69-2520 Fax: +49 3677 69-1743 e-mail: kongressorganisation@tu-ilmenau.de Redaktionsschluss: Juli 2007 Verlag: Technische Universität Ilmenau/Universitätsbibliothek

Universitätsverlag Ilmenau Postfach 10 05 65 98684 Ilmenau www.tu-ilmenau.de/universitaetsverlag

Herstellung und Verlagshaus Monsenstein und Vannerdat OHG Auslieferung: Am Hawerkamp 31 48155 Münster www.mv-verlag.de Layout Cover: www.cey-x.de

Bezugsmöglichkeiten: Universitätsbibliothek der TU Ilmenau Tel.: +49 3677 69-4615 Fax: +49 3677 69-4602

© Technische Universität Ilmenau (Thür.) 2007 Diese Publikationen und alle in ihr enthaltenen Beiträge und Abbildungen sind urheberrechtlich geschützt. Mit Ausnahme der gesetzlich zugelassenen Fälle ist eine Verwertung ohne Einwilligung der Redaktion strafbar.

Preface Dear Participants, Confronted with the ever-increasing complexity of technical processes and the growing demands on their efficiency, security and flexibility, the scientific world needs to establish new methods of engineering design and new methods of systems operation. The factors likely to affect the design of the smart systems of the future will doubtless include the following:

• As computational costs decrease, it will be possible to apply more complex algorithms, even in real time. These algorithms will take into account system nonlinearities or provide online optimisation of the system’s performance.

• New fields of application will be addressed. Interest is now being expressed, beyond that in “classical” technical systems and processes, in environmental systems or medical and bioengineering applications.

• The boundaries between software and hardware design are being eroded. New design methods will include co-design of software and hardware and even of sensor and actuator components.

• Automation will not only replace human operators but will assist, support and supervise humans so that their work is safe and even more effective.

• Networked systems or swarms will be crucial, requiring improvement of the communication within them and study of how their behaviour can be made globally consistent.

• The issues of security and safety, not only during the operation of systems but also in the course of their design, will continue to increase in importance.

The title “Computer Science meets Automation”, borne by the 52nd International Scientific Colloquium (IWK) at the Technische Universität Ilmenau, Germany, expresses the desire of scientists and engineers to rise to these challenges, cooperating closely on innovative methods in the two disciplines of computer science and automation.

The IWK has a long tradition going back as far as 1953. In the years before 1989, a major function of the colloquium was to bring together scientists from both sides of the Iron Curtain. Naturally, bonds were also deepened between the countries from the East. Today, the objective of the colloquium is still to bring researchers together. They come from the eastern and western member states of the European Union, and, indeed, from all over the world. All who wish to share their ideas on the points where “Computer Science meets Automation” are addressed by this colloquium at the Technische Universität Ilmenau. All the University’s Faculties have joined forces to ensure that nothing is left out. Control engineering, information science, cybernetics, communication technology and systems engineering – for all of these and their applications (ranging from biological systems to heavy engineering), the issues are being covered. Together with all the organizers I should like to thank you for your contributions to the conference, ensuring, as they do, a most interesting colloquium programme of an interdisciplinary nature. I am looking forward to an inspiring colloquium. It promises to be a fine platform for you to present your research, to address new concepts and to meet colleagues in Ilmenau. Professor Peter Scharff Professor Christoph Ament Rector, TU Ilmenau Head of Organisation

Ta

ble

of

Co

nte

nts

III

C O N T E N T S Page 1 Systems Engineering and Intelligent Systems

A. Yu. Nedelina, W. Fengler 3 DIPLAN: Distributed Planner for Decision Support Systems

O. Sokolov, M. Wagenknecht, U. Gocht 9 Multiagent Intelligent Diagnostics of Arising Faults

V. Nissen 15 Management Applications of Fuzzy Conrol

O. G. Rudenko, A. A. Bessonov, P. Otto 21 A Method for Information Coding in CMAC Networks

Ye. Bodyanskiy, P. Otto, I. Pliss, N. Teslenko 27 Nonlinear process identification and modeling using general regression neuro-fuzzy network

Ye. Bodyanskiy, Ye. Gorshkov, V. Kolodyazhniy , P. Otto 35 Evolving Network Based on Double Neo-Fuzzy Neurons

Ch. Wachten, Ch. Ament, C. Müller, H. Reinecke 41 Modeling of a Laser Tracker System with Galvanometer Scanner

K. Lüttkopf, M. Abel, B. Eylert 47 Statistics of the truck activity on German Motorways

K. Meissner, H. Hensel 53 A 3D process information display to visualize complex process conditions in the process industry

F.-F. Steege, C. Martin, H.-M. Groß 59 Recent Advances in the Estimation of Pointing Poses on Monocular Images for Human-Robot Interaction

A. González, H. Fernlund, J. Ekblad 65 After Action Review by Comparison – an Approach to Automatically Evaluating Trainee Performance in Training Exercise

R. Suzuki, N. Fujiki, Y. Taru, N. Kobayashi, E. P. Hofer 71 Internal Model Control for Assistive Devices in Rehabilitation Technology

D. Sommer, M. Golz 77 Feature Reduction for Microsleep Detection

IV

F. Müller, A. Wenzel, J. Wernstedt 83 A new strategy for on-line Monitoring and Competence Assignment to Driver and Vehicle

V. Borikov 89 Linear Parameter-Oriented Model of Microplasma Process in Electrolyte Solutions

A. Avshalumov, G. Filaretov 95 Detection and Analysis of Impulse Point Sequences on Correlated Disturbance Phone

H. Salzwedel 101 Complex Systems Design Automation in the Presence of Bounded and Statistical Uncertainties

G. J. Nalepa, I. Wojnicki 107 Filling the Semantic Gaps in Systems Engineering

R. Knauf 113 Compiling Experience into Knowledge

R. Knauf, S. Tsuruta, Y. Sakurai 119 Toward Knowledge Engineering with Didactic Knowledge 2 Advances in Control Theory and Control Engineering

U. Konigorski, A. López 129 Output Coupling by Dynamic Output Feedback

H. Toossian Shandiz, A. Hajipoor 135 Chaos in the Fractional Order Chua System and its Control

O. Katernoga, V. Popov, A. Potapovich, G. Davydau 141 Methods for Stability Analysis of Nonlinear Control Systems with Time Delay for Application in Automatic Devices

J. Zimmermann, O. Sawodny 145 Modelling and Control of a X-Y-Fine-Positioning Table

A. Winkler, J. Suchý 151 Position Based Force Control of an Industrial Manipulator

E. Arnold, J. Neupert, O. Sawodny, K. Schneider 157 Trajectory Tracking for Boom Cranes Based on Nonlinear Control and Optimal Trajectory Generation

V

K. Shaposhnikov, V. Astakhov 165 The method of ortogonal projections in problems of the stationary magnetic field computation

J. Naumenko 167 The computing of sinusoidal magnetic fields in presence of the surface with bounded conductivity

K. Bayramkulov, V. Astakhov 169 The method of the boundary equations in problems of computing static and stationary fields on the topological graph

T. Kochubey, V. Astakhov 171 The computation of magnetic field in the presence of ideal conductors using the Integral-differential equation of the first kind M. Schneider, U. Lehmann, J. Krone, P. Langbein, Ch. Ament, P. Otto, 173 U. Stark, J. Schrickel Artificial neural network for product-accompanied analysis and control

I. Jawish 179 The Improvement of Traveling Responses of a Subway Train using Fuzzy Logic Techniques

Y. Gu, H. Su, J. Chu 185 An Approach for Transforming Nonlinear System Modeled by the Feedforward Neural Networks to Discrete Uncertain Linear System 3 Optimisation and Management of Complex Systems and Networked Systems

R. Franke, J. Doppelhammer 193 Advanced model based control in the Industrial IT System 800xA

H. Gerbracht, P. Li, W. Hong 199 An efficient optimization approach to optimal control of large-scale processes

T. N. Pham, B. Wutke 205 Modifying the Bellman’s dynamic programming to the solution of the discrete multi-criteria optimization problem under fuzziness in long-term planning

S. Ritter, P. Bretschneider 211 Optimale Planung und Betriebsführung der Energieversorgung im liberalisierten Energiemarkt

P. Bretschneider, D. Westermann 217 Intelligente Energiesysteme: Chancen und Potentiale von IuK-Technologien

VI

Z. Lu, Y. Zhong, Yu. Wu, J. Wu 223 WSReMS: A Novel WSDM-based System Resource Management Scheme

M. Heit, E. Jennenchen, V. Kruglyak, D. Westermann 229 Simulation des Strommarktes unter Verwendung von Petrinetzen

O. Sauer, M. Ebel 237 Engineering of production monitoring & control systems

C. Behn, K. Zimmermann 245 Biologically inspired Locomotion Systems and Adaptive Control

J. W. Vervoorst, T. Kopfstedt 251 Mission Planning for UAV Swarms

M. Kaufmann, G. Bretthauer 257 Development and composition of control logic networks for distributed mechatronic systems in a heterogeneous architecture

T. Kopfstedt, J. W. Vervoorst 263 Formation Control for Groups of Mobile Robots Using a Hierarchical Controller Structure

M. Abel, Th. Lohfelder 269 Simulation of the Communication Behaviour of the German Toll System

P. Hilgers, Ch. Ament 275 Control in Digital Sensor-Actuator-Networks

C. Saul, A. Mitschele-Thiel, A. Diab, M. Abd rabou Kalil 281 A Survey of MAC Protocols in Wireless Sensor Networks

T. Rossbach, M. Götze, A. Schreiber, M. Eifart, W. Kattanek 287 Wireless Sensor Networks at their Limits – Design Considerations and Prototype Experiments

Y. Zhong, J. Ma 293 Ring Domain-Based Key Management in Wireless Sensor Network

V. Nissen 299 Automatic Forecast Model Selection in SAP Business Information Warehouse under Noise Conditions M. Kühn, F. Richter, H. Salzwedel 305 Process simulation for significant efficiency gains in clinical departments – practical example of a cancer clinic

VII

D. Westermann, M. Kratz, St. Kümmerling, P. Meyer 311 Architektur eines Simulators für Energie-, Informations- und Kommunikations- technologien P. Moreno, D. Westermann, P. Müller, F. Büchner 317 Einsatzoptimierung von dezentralen netzgekoppelten Stromerzeugungs- anlagen (DEA) in Verteilnetzen durch Erhöhung des Automatisierungsgrades

M. Heit, S. Rozhenko, M. Kryvenka, D. Westermann 331 Mathematische Bewertung von Engpass-Situationen in Transportnetzen elektrischer Energie mittels lastflussbasierter Auktion

M. Lemmel, M. Schnatmeyer 339 RFID-Technology in Warehouse Logistics

V. Krugljak, M. Heit, D. Westermann 345 Approaches for modelling power market: A Comparison.

St. Kümmerling, N. Döring, A. Friedemann, M. Kratz, D. Westermann 351 Demand-Side-Management in Privathaushalten – Der eBox-Ansatz 4 Intelligent Vehicles and Mobile Systems

A. P. Aguiar, R. Ghabchelloo, A. Pascoal, C. Silvestre , F. Vanni 359 Coordinated Path following of Multiple Marine Vehicles: Theoretical Issues and Practical Constraints

R. Engel, J. Kalwa 365 Robust Relative Positioning of Multiple Underwater Vehicles

M. Jacobi, T. Pfützenreuter, T. Glotzbach, M. Schneider 371 A 3D Simulation and Visualisation Environment for Unmanned Vehicles in Underwater Scenarios

M. Schneider, M. Eichhorn, T. Glotzbach, P. Otto 377 A High-Level Simulator for heterogeneous marine vehicle teams under real constraints

A. Zangrilli, A. Picini 383 Unmanned Marine Vehicles working in cooperation: market trends and technological requirements

T. Glotzbach, P. Otto, M. Schneider, M. Marinov 389 A Concept for Team-Orientated Mission Planning and Formal Language Verification for Heterogeneous Unmanned Vehicles

VIII

M. A. Arredondo, A. Cormack 395 SeeTrack: Situation Awareness Tool for Heterogeneous Vehicles

J. C. Ferreira, P. B. Maia, A. Lucia, A. I. Zapaniotis 401 Virtual Prototyping of an Innovative Urban Vehicle

A. Wenzel, A. Gehr, T. Glotzbach, F. Müller 407 Superfour-in: An all-terrain wheelchair with monitoring possibilities to enhance the life quality of people with walking disability

Th. Krause, P. Protzel 413 Verteiltes, dynamisches Antriebssystem zur Steuerung eines Luftschiffes T. Behrmann, M. Lemmel 419 Vehicle with pure electric hybrid energy storage system

Ch. Schröter, M. Höchemer, H.-M. Groß 425 A Particle Filter for the Dynamic Window Approach to Mobile Robot Control

M. Schenderlein, K. Debes, A. Koenig, H.-M. Groß 431 Appearance-based Visual Localisation in Outdoor Environments with an Omnidirectional Camera

G. Al Zeer, A. Nabout, B. Tibken 437 Hindernisvermeidung für Mobile Roboter mittels Ausweichecken 5 Robotics and Motion Systems

Ch. Schröter, H.-M. Groß 445 Efficient Gridmaps for SLAM with Rao-Blackwellized Particle Filters

St. Müller, A. Scheidig, A. Ober, H.-M. Groß 451 Making Mobile Robots Smarter by Probabilistic User Modeling and Tracking

A. Swerdlow, T. Machmer, K. Kroschel, A. Laubenheimer, S. Richter 457 Opto-acoustical Scene Analysis for a Humanoid Robot

A. Ahranovich, S. Karpovich, K. Zimmermann 463 Multicoordinate Positioning System Design and Simulation

A. Balkovoy, V. Cacenkin, G. Slivinskaia 469 Statical and dynamical accuracy of direct drive servo systems

Y. Litvinov, S. Karpovich, A. Ahranovich 477 The 6-DOF Spatial Parallel Mechanism Control System Computer Simulation

V. Lysenko, W. Mintchenya, K. Zimmermann 483 Minimization of the number of actuators in legged robots using biological objects

J. Kroneis, T. Gastauer, S. Liu, B. Sauer 489 Flexible modeling and vibration analysis of a parallel robot with numerical and analytical methods for the purpose of active vibration damping

A. Amthor, T. Hausotte, G. Jäger, P. Li 495 Friction Modeling on Nanometerscale and Experimental Verification Paper submitted after copy deadline 2 Advances in Control Theory and Control Engineering V. Piwek, B. Kuhfuss, S. Allers Feed drivers – Synchronized Motion is leading to a process optimization 503

IX

52nd Internationales Wissenschaftliches Kolloquium Technische Universität Ilmenau

10 – 13 September 2007

J. W. Vervoorst / T. Kopfstedt

Mission Planning for UAV Swarms

Introduction

Unmanned Aerial Vehicles (UAVs) play an ever increasing role in a wide variety of

scenarios in both the civilian and military sector, carrying out tasks like traffic

surveillance, firefighting, or reconnaissance missions. Furthermore, the use of groups of

vehicles, or swarms, has been shown to accomplish certain objectives more efficiently

and more effectively than a single vehicle can, for example in terrain mapping or search

missions. In all these scenarios, the autonomous vehicles need to fly on trajectories that

match their flight envelope, are as short as possible, and most importantly, avoid

collisions with obstacles and other UAVs at all cost.

In order to achieve this, Mixed-Integer Linear Programming (MILP) is used as the

optimization principle. MILP extends regular linear programming to include variables that

are constrained to integer or binary values. Thus, MILP offers the possibility to add

logical and decision making constraints into the optimization, such as obstacle and

collision avoidance.

However, finding long-range minimum-time trajectories in environments with many

obstacles is a complex optimization problem. In this paper, a Model Predictive Control

(MPC) approach is chosen to decrease computational complexity and limit computation

time, therefore making the algorithm capable of real-time calculations as well as

handling unknown or dynamically changing environments. The presented algorithm is

capable of calculating near-optimal flight trajectories to ensure that the UAV swarm

carries out its mission in the minimum time.

Problem Formulation

In general, a mission for a UAV swarm can be described as visiting a number of

waypoints spread out on a map. All waypoints have to be visited once during the

mission by one UAV. This so called Task Assignment constitutes the first part of the

mission planning algorithm. During the Task Assignment phase, waypoints are assigned

such that each UAV will have to travel only a minimum distance, therefore also bringing

251

the total mission time to a minimum. The problem is formulated as a multi-dimensional

Traveling Salesman Problem (TSP), as implemented in [1].

In this paper, only planar motion is considered, meaning UAVs fly at constant altitude.

The dynamics of the UAVs are modeled in the form of a simple point mass, bringing

their discrete-time state space representation to:

( )

( )

ky

x

ky

x

ky

x a

a

t

t

t

t

v

v

y

x

t

t

v

v

y

x

⋅

∆

∆

∆

∆

+

⋅

∆

∆

=

+0

0

2/0

02/

1000

0100

010

001

2

2

1

UAV motion is constrained by a maximum velocity and turn rate, therefore giving us:

maxmax , vvvv yx ≤≤ maxmax , aaaa yx ≤≤

Calculating the length of a vector is a nonlinear operation. In order to adhere to the

Linear Programming problem formulation, we approximate vector length by checking a

number maxvn of unit vectors onto which the velocity/acceleration vector is projected [5].

max,...,1 max vk

Tnkviv =≤⋅

max,...,1 max ak

Tnkaia =≤⋅ (1)

=

max

max

2sin

2cos

v

v

k

n

k

n

k

iπ

π

=

max

max

2sin

2cos

a

a

k

n

k

n

k

iπ

π

(2)

MILP also allows an efficient way to declare obstacles by using binary variables.

Throughout this paper we limit ourselves to rectangular obstacles. Obstacles are

described by their lower left and upper right corner [ ]Tururllll yxyx ,,, . The variable objectb

is a binary, and the number M is an arbitrary large positive value.

( )

( )

( )

( )4

3

2

1

ijk

ijk

ijk

ijk

objecturjk

objecturjk

objectlljk

objectlljk

bMxy

bMxx

bMxy

bMxx

⋅−≥

⋅−≥

⋅+≤

⋅+≤

4...1 , ...1

...1 , ...1

3

O

4

1

==

==

≤∑=

lTk

NjNi

b

V

l

objectijkl

(3)

[ ]T

jkjk yx

, is the position of UAV j at time k . Index i describes the number of obstacles,

j lists the number of UAVs in the swarm, k the number of time steps in the planning

horizon, and l enumerates the four inequalities. As long as 3≤∑ ijklobjectb is fulfilled, the

UAV is outside the obstacle.

252

Receding Horizon Controller

The constraints above can now be used to formulate a MILP optimization problem.

Traditionally, trajectory optimization is done over a fixed horizon. This means that the

complete trajectory is calculated from beginning to end point, forming a large and

complicated optimization that does not take into account changing environments. When

obstacles are added or subtracted, the precalculated optimal solution essentially

becomes worthless and a recalculation has to be performed.

To solve this problem, a Model Predictive Control setup is chosen. [2] lists the properties

of Model Predictive Control, also called Receding Horizon Control:

• At time i and initial state ix the optimization is performed for only the next PN

time steps. PN is the so called planning horizon.

• The first EN values of the optimal solution are used as inputs to the system. EN

is the execution horizon. EN is usually set to 1.

• In the new initial state ENix + the optimization is performed again, repeating the

process, thus taking into account possible changes in the environment.

Similar implementations of MPC can be found in [3] and [4].

To be able to use this approach, we need to overcome a problem. As seen in Fig. 1, the

trajectory is only optimized within a small area around the current state, the planning

horizon. But the total trajectory consists of the optimized piece plus the remaining pieces

from the planning horizon to the goal point. However, that piece is unknown and the

total distance cannot be calculated. Therefore, the remaining trajectory is approximated

with straight line segments connecting the planning horizon to the goal point.

Fig. 1: Details of MPC Fig. 2: Calculation of the cost map

253

In order to find the shortest distance to the goal point, a cost map of the complete

environment is calculated, calculating and storing the distances of points to the goal

point (Fig. 2). [6] shows that the shortest distance in an environment with convex

obstacles is a combination of line segments between the points and the corners of the

obstacles. Thus, the corners of the obstacles are cost points in our cost map. The

visibility graph between all obstacles and the goal point is calculated and by using the

Dijkstra algorithm, as outlined in [7], the shortest connections between all the points and

the goal point are calculated and stored as the cost of each point.

The length of the trajectory outside the planning horizon can now be determined. As

detailed in Fig. 1, it is:

• The distance from PNix + ,the edge of the planning horizon, to a known cost point

optx that is visible from that point.

• The distance from that cost point to the goal point. This value has already been

calculated and is stored in the cost map.

The visibility constraint between PNix + and optx is very important. Because of it, local

optimization within the planning horizon does indeed have a global influence on the

whole trajectory, therefore also minimizing the total trajectory length.

Now the planning problem is expressed in MILP form. Each UAV can select only one

cost point during each optimization:

V

N

i

N

j

CP NkbCP goal

ijk,...,1 , 1

1 1

==∑ ∑= =

with CPN = number of cost points, VN = number of UAVs, goalN =2, the next two waypoints.

Equations (4) and (5) set the cost value and coordinates of the chosen cost point.

V

N

i

N

j

CPiopt NkbccCP goal

ijkk,...,1 ,

1 1

=⋅= ∑ ∑= =

(4)

V

N

i

N

j

CPCP

CP

opt

optNkb

y

x

y

x CP goal

ijk

k

k

k

k,...,1 ,

1 1

=⋅

=

∑ ∑

= =

(5)

The line connecting the edge of the planning horizon and the cost point is described by

( )( )

−

=

+

+

kNi

kNi

opt

opt

line

line

P

P

k

k

k

k

y

x

y

x

y

x

In order to test visibility, this connection is divided into testN parts:

254

( )( ) test

line

line

testkNi

kNi

test

testNm

y

x

N

m

y

x

y

x

k

k

P

P

km

km...1 =

⋅+

=

+

+

Each part is tested for interference with obstacles using a binary variable ikmnoptb , very

much like in (3).

Equations (6) and (7) handle which goal point to use in the optimization. If a goal point

is reached in the last step of the planning horizon or not at all, then only this point will be

part of the optimization. If a goal point is reached within the planning horizon, then goal

points are switched and the optimization directs the trajectory to the next goal point.

∑∑+

==

=1

11

T

Ti

goal

N

i

CP ij

CP

kibb (6)

∑∑−

==

=1

112

T

i

goal

N

i

CP ij

CP

kibb (7)

Cost Function

The cost function to be minimized is the total trajectory lengths of the UAVs, consisting

of three parts:

• The part within the planning horizon, from ix to PNix +

• The line between PNix + and the selected cost point

• The distance between selected cost point and goal point.

The first part is described by the first part of Equation (8), which represents the number

of time steps to the goal point times the distance traveled per time step.

( )∑ ∑=

+

++⋅⋅∆⋅=

V

jkj

N

j

optline

T

k

goalj clbktvZ1

1

max (8)

linel is the distance between PNix + and the selected cost point. Since it constitutes the

length of a vector, it is calculated just as in Equations (1) and (2).

The third part is simply the stored value of the cost point, meaning the distance to the

goal point.

Simulation

Two scenarios are presented that show the capabilities of the presented MCD algorithm,

especially in changing environments. Fig. 3 shows a trajectory replanning after a UAV is

lost during a mission.

255

Fig. 3: Replanning after UAV loss Fig. 4: Replanning due to unknown obstacles

The regularly planned trajectories are shown in black. UAV #2 is supposed to visit

waypoints 5

W and 6

W . However, UAV #2 is lost at 5

W . Immediately, a replanning occurs

and UAV #3 takes over 6

W . The newly planned trajectories are shown in gray.

In Fig. 4, UAV #1 is scheduled to visit 1

W and 2

W . However, it encounters an unknown

obstacles, shown in gray, that was not part of the previous planning. After a replanning,

it is determined that UAV #2 can reach 2

W faster and it takes over for #1.

Conclusion

The iterative MPC algorithm presented here has been shown to effectively handle

various changes in the mission environment, decreasing computational complexity and

still being able to calculate near-optimal trajectories for groups of cooperating UAVs.

References: [1] SCHUMACHER, Corey; CHANDLER, Phillip; PACHTER, Meir: Constrained Optimization for UAV Task Assignment. AIAA Guidance, Navigation, and Control Conference, Providence, RI, USA, 16-19 August 2004 [2] MACIEJOWSKI, Jan M.: Predictive Control with Constraints. London: Prentice Hall, 2002 [3] RICHARDS, Arthur; HOW, Jonathan P.: Model Predictive Control of Vehicle Maneuvers with Guaranteed Completion Time and Robust Feasibility. American Control Conference, Denver, CO, USA, 4-6 June 2003 [4] FREW, Eric W.: Receding Horizon Control Using Random Search for UAV Navigation with Passive, Non-cooperative Sensing. AIAA Guidance, Navigation, and Control Conference, San Francisco, CA, USA, August 2005 [5] RICHARDS, Arthur; HOW, Jonathan P.: Mixed-integer Programming for Control. American Control Conference, pp. 2676- 2683, June 8-10, 2005 [6] LATOMBE, J.-C.: Robot Motion Planning. Boston, MA, USA: Kluwer Academic Publishers, 1991, pp. 153-200 [7] CORMEN, Thomas H.: Introduction to Algorithms. Cambridge, MA, USA: The MIT Press, 2001, pp. 595-601, 629-635

Authors: Dipl.-Ing. Jan Willem Vervoorst Dipl.-Ing. Thomas Kopfstedt Diehl BGT Defence GmbH & Co. KG, Alte Nußdorfer Str. 13 D-88662 Überlingen, Germany Phone: +49 7551 89-2449 Fax: +49 7551 89-2351 E-mail: jan.vervoorst@diehl-bgt-defence.de

256